During the soldering process of high-frequency boards, voids inside solder joints are not only a sign of cold or poor soldering, but also a key cause of nonlinear fluctuations in dielectric constant and loss tangent. Such fluctuations directly amplify the dispersion of S-parameters in frequency bands above 28GHz, resulting in out-of-tolerance whole-machine link budgets. We once found in the mass production of a certain millimeter-wave radar PCB that when the void rate in the same LOT rose from 2.1% to 5.8%, the -3dB bandwidth offset reached 1.7GHz, and this offset could not be compensated by post-calibration.

Void formation mechanism is strongly correlated with Solder Paste formulation

Voids are not solely determined by reflow atmosphere or temperature profile; their root cause lies in the matching relationship among the particle size distribution of metal powder in the solder paste, the thermal decomposition onset temperature of Flux resin, and the gas escape rate. Jia Jinyuan low-void solder paste strictly controls the D50 of Sn96.5Ag3.0Cu0.5 alloy powder at 18.3μm±0.4μm, while setting the main decomposition peak temperature of the rosin-based flux at 142℃, which falls exactly within the window after solder paste collapse is completed and before the liquidus temperature (217℃), thereby preventing gas from being trapped and locked by liquid solder.

Key points of coordinated control in printing and placement processes

The stencil aperture size must be dynamically adapted to the solder paste particle size. When using a 0.08mm thick laser-etched stencil, if the aperture width is too small, the squeegee pressure must be increased synchronously to 5.2kg to ensure filling rate, but excessive pressure will exacerbate flux extrusion and induce edge voids. Therefore, we adopt a stepped squeegee angle design: using a 55° angle in the initial printing stage to achieve initial filling, and automatically switching to a 62° angle in the latter half to complete compaction. This approach reduces the occurrence rate of voids at the pad edges of QFN devices by 41%.

Reflow process must avoid three types of hidden risk points

Although an oxygen content below 100ppm in a nitrogen atmosphere is beneficial for reducing oxidation, if the dew point is higher than -40℃, microbubbles generated by the reaction of water vapor with flux residues are instead more difficult to discharge. Therefore, the regeneration temperature of the nitrogen dryer must be locked at 180℃, and the adsorption cycle set to 4.5 minutes. In addition, insufficient time in the soaking zone will lead to incomplete activation of organic acids, while excessive time will cause carbonization and retention of some resins. In actual production lines, we set the 150–180℃ interval to 98 seconds, with an upper and lower fluctuation of no more than ±3 seconds. Finally, if the peak temperature exceeds 245℃, although it can reduce the void rate, it will accelerate the excessive growth of Cu6Sn5 intermetallic compounds, shortening the high-frequency mechanical fatigue life of solder joints by 37%.

The judgment boundaries of AOI and X-ray inspection must be redefined

Traditional AOI only alarms for voids based on area proportion, but high-frequency boards should incorporate position weight judgment. For example, a single void with a diameter >35μm in the central area of a BGA must be intercepted, while edge solder joints are allowed to have multiple voids <20μm, as long as they are not connected and are >8μm away from the pad boundary. In X-ray images, if voids are regularly circular with sharp edges, they are most likely caused by flux volatilization; if they are torn or burr-shaped, they point to stencil wear or placement offset. In such cases, the line must be stopped immediately to replace the stencil and retest SPI data.